898 METALLURGY OF IRON AND STEEL.
letter of the lam In steel above .08 per cent, of phosphorus, this difference in cost disappears.
SEC. XVIIId.—Tests on plates.—A spread of 10,000 pounds per square inch in the ultimate strength should he allowed on all sections, but it is especially necessary on plates. In trying to fill rigid specifications where no allowance is made for thickness, or where the allowable limits of strength are too narrow, the plate rollers have been driven to expedients which are dangerously near the line of deception. Thus, if it is required that a test be cut from one plate out of ten, the manufacturer will leave a coupon on every plate and test strips are cut from immediately next to them; after finding which plates fill the requirements, the coupons are cut from the others and the inspector is told that the pile is ready for him.
If every plate is to be tested, then a coupon is left upon each corner and a contiguous strip is privately tested by the maker. After finding which corner gives the best results, the other coupons are cut off and the plate submitted to the inspector. This is not dishonest, for any one corner represents the plate just as much as any other corner, and it would manifestly be absurd to designate from which corner the test is to be taken. It is also certain that no one corner represents the center of the plate, for the edges are finished colder than the center, and in a plate rolled direct from an ingot the corners in no way represent the part which corresponds to the segregated portion of the ingot.
It is by care in the preliminary testing rather than by improvement in the quality of material that advances have been made. The mill managers have been aided by the inspectors, for most of these men are anxious to pass material which they know to be good. They allow the manufacturer to put part of a heat into thick plates and part into thin, and make the tests on three-eighths or one-half inch gauge; they pass over the sheets that are 100 inches wide, and cut the coupons from plates that are less than 70 inches. On the other hand, higher tests should be called for on plates under 4£ inches wide. This is because they can be made on a universal mill, and since better results can be had in this way, it is right to demand what there is a simple way of obtaining. No allowance need be made for a variation in tensile strength for different shapes, but concessions should be made for differences in thickness. This arises from the fact that it is generally known beforehand whether a cer-